Yes, I can confirm the problem. It did not happen with the full script that you posted, but it happened with the script from the zip, and is indeed caused by timeOffset. Since I can not see the problem in the source code, I've forwarded this to the developers for checking. Thank you for finding this bug!

Update: The bug was supposedly fixed in version 1.50.4. Please check if there's still a problem with this version. This bug was also responsible for the wrong day numbers of ldow() with time zones.